NEW YORK (GenomeWeb) – Single-cell sequencing can help reveal karyotype heterogeneity and may someday inform cancer treatment approaches, according to researchers led by the University of Groningen in the Netherlands. Groningen researcher Floris Foijer and his colleagues turned to whole-genome single-cell sequencing to examine the aneuploidy that occurs in hematological cancers and whether the similar karyotypes they observed in a mouse model of lymphoma were due to the cancer cells overcoming the pitfalls of chromosomal instability or to selection of cells with a particular karyotype.
As the group reported in Genome Biology this week, it's a bit of both. Using single-cell sequencing and an in-house algorithm they developed to determine chromosome copy numbers from the data , they found that aneuploid T-cell lymphoma mouse models and human B-cell leukemias exhibit both copy number heterogeneity as well as specific and recurring chromosomal changes. As such chromosomal instability — and thus tumor heterogeneity — could drive tumor evolution, the researchers said that single cell-based karyotype analysis could become a key tool in determining cancer treatment approaches.
"This heterogeneity is something that we can now measure, and we now have a tool to measure it," Foijer told GenomeWeb. "We are now going to test how predictive this is going to be for disease outcome."
As a postdoc, Foijer developed a mouse model of aneuploid cancer by inactivating crucial chromosome segregation genes and then crossing those mice with a cancer-prone line with inactivated p53 genes. The resulting mice were even more disposed to developing cancer. Foijer expected that chromosome segregation would be affected in these tumors and that he'd observe randomly re-segregated chromosomes.
That wasn't the case, though. While the mouse tumors were extremely aneuploid, they also followed a certain pattern. By array CGH, he and his colleagues found that the tumors all had extra copies of chromosomes 4, 14, and 15.
When they went to publish what they'd found two years ago, reviewers noted that this pattern could be due to either temporary chromosomal instability followed by tumor evolution toward a stable karyotype, resulting in an aneuploid tumor with stable chromosomes, or due to evolutionary pressure toward the development of particular karyotypes, which would also result in this karyotype, but with some random re-segregation on other chromosomes that an approach like array CHG, which captures bulk data on chromosomal abnormalities, wouldn't pick up.
"The only real method that we could think of to quantify every single chromosome in a dividing or non-dividing cell in the primary tumor was by sequencing the individual cells," Foijer said.
For their current study, they sequenced the genomes of tumor cells they isolated from the mouse models. The bulk sequencing data largely reflected the array CGH data they'd collected.
To annotate the karyotypes of each individual cell, Foijer and his colleagues developed an analysis pipeline called AneuFinder that bins mapped reads and uses a Hidden Markov Model to gauge the copy number status of each bin. This software is available via Bioconductor, they noted in their paper.
Through single-cell sequencing of 48 cells from each mouse model of cancer, the researchers found that all had clonal chromosomal copy number gains, as they'd found through array CGH. However, they also uncovered other chromosomal copy number variations in a minority of lymphoma cells. These changes, they noted, weren't detected when they analyzed the single-cell sequencing reads in bulk, and wouldn't have been caught by array CGH.
This also suggested to them that their mouse models are highly aneuploid and exhibit high-grade karyotype heterogeneity, which in turn suggests there's ongoing chromosomal instability in these tumors.
He and his colleagues then applied this approach to examine the karyotypes of three human B-cell leukemia samples with varying levels of aneuploidy, as gauged through traditional cytogenetic approaches. All of these samples exhibited different heterogeneity rates, Foijer said.
The one with the highest level of aneuploidy also exhibited karyotype heterogeneity. This sample was nearly triploid, with a number of whole chromosome and local copy number alterations.
This indicated to Foijer that heterogeneity could be important for treatment. "If you've got a high mis-segregation rate in a tumor that will result in this heterogeneity, then this tumor is probably better at evolving further when it is pushed into another environment, which would be treatment," he said.
To test this, he and his colleagues grafted this human tumor into immune-deficient mice. Indeed, they found that its karyotype changed after transplantation, as a smattering of cells gained a copy of chromosome 2 and lost the extra copy of chromosome 9 that they had had, hinting that it could be better at handling such new situations.
In an accompanying Genome Biology paper, Foijer and a team led by Peter Lansdorp at the BC Cancer Agency in Canada used the same single-cell sequencing approach to evaluate the karyotypes of frontal cortex neurons from people with and without Alzheimer's disease. Previous interphase FISH-based studies had suggested that brain cells from patients with Alzheimer's have an increased number of aneuploid cells. But with single-cell sequencing and AneuFinder, the researchers found this is likely not the case.
While this suggests that their approach could be applied to study aneuploidy in other diseases, Foijer noted that its mostly common application would likely be in cancer, as more than two thirds oftumorsare aneuploid.
But to be used in the clinic, Foijer said, he must first convince physicians of its usefulness, likely by linking karyotype heterogeneity to disease outcome or to treatment choices.
He noted that a number of cancer therapies are aimed at making cancer cells aneuploid and thus less stable. Treating an already aneuploid or chromosomally unstable cancer with such a drug could have one or two outcomes: it could either push that cancer over the edge so it'll die or it could have a limited effect as the cells are already able to cope with instability and perhaps should be treated with a different therapy instead.
"We know now that the heterogeneity exists and we are quite confident that it will somehow be predictive, but we need to show that first," Foijer said.
Foijer is lining up national and international collaborators to do just that. "We are now looking in many more human tumors to actually see whether there is a link between heterogeneity and outcomes," he said.